Tricks

$Tricks$

---

• 1.mosaic

def load_mosaic(self, index):
    # loads images in a mosaic

    labels4 = []
    s = self.img_size
    yc, xc = [int(random.uniform(-x, 2 * s + x)) for x in self.mosaic_border]  # mosaic center x, y
    indices = [index] + [random.randint(0, len(self.labels) - 1) for _ in range(3)]  # 3 additional image indices
    for i, index in enumerate(indices):
        # Load image
        img, _, (h, w) = load_image(self, index)

        # place img in img4
        if i == 0:  # top left
            img4 = np.full((s * 2, s * 2, img.shape[2]), 114, dtype=np.uint8)  # base image with 4 tiles
            x1a, y1a, x2a, y2a = max(xc - w, 0), max(yc - h, 0), xc, yc  # xmin, ymin, xmax, ymax (large image)
            x1b, y1b, x2b, y2b = w - (x2a - x1a), h - (y2a - y1a), w, h  # xmin, ymin, xmax, ymax (small image)
        elif i == 1:  # top right
            x1a, y1a, x2a, y2a = xc, max(yc - h, 0), min(xc + w, s * 2), yc
            x1b, y1b, x2b, y2b = 0, h - (y2a - y1a), min(w, x2a - x1a), h
        elif i == 2:  # bottom left
            x1a, y1a, x2a, y2a = max(xc - w, 0), yc, xc, min(s * 2, yc + h)
            x1b, y1b, x2b, y2b = w - (x2a - x1a), 0, max(xc, w), min(y2a - y1a, h)
        elif i == 3:  # bottom right
            x1a, y1a, x2a, y2a = xc, yc, min(xc + w, s * 2), min(s * 2, yc + h)
            x1b, y1b, x2b, y2b = 0, 0, min(w, x2a - x1a), min(y2a - y1a, h)

        img4[y1a:y2a, x1a:x2a] = img[y1b:y2b, x1b:x2b]  # img4[ymin:ymax, xmin:xmax]
        padw = x1a - x1b
        padh = y1a - y1b

        # Labels
        x = self.labels[index]
        labels = x.copy()
        if x.size > 0:  # Normalized xywh to pixel xyxy format
            labels[:, 1] = w * (x[:, 1] - x[:, 3] / 2) + padw
            labels[:, 2] = h * (x[:, 2] - x[:, 4] / 2) + padh
            labels[:, 3] = w * (x[:, 1] + x[:, 3] / 2) + padw
            labels[:, 4] = h * (x[:, 2] + x[:, 4] / 2) + padh
        labels4.append(labels)

    # Concat/clip labels
    if len(labels4):
        labels4 = np.concatenate(labels4, 0)
        # np.clip(labels4[:, 1:] - s / 2, 0, s, out=labels4[:, 1:])  # use with center crop
        np.clip(labels4[:, 1:], 0, 2 * s, out=labels4[:, 1:])  # use with random_affine

    # Augment
    # img4 = img4[s // 2: int(s * 1.5), s // 2:int(s * 1.5)]  # center crop (WARNING, requires box pruning)
    img4, labels4 = random_affine(img4, labels4,
                                  degrees=self.hyp['degrees'],
                                  translate=self.hyp['translate'],
                                  scale=self.hyp['scale'],
                                  shear=self.hyp['shear'],
                                  border=self.mosaic_border)  # border to remove

    return img4, labels4

---

• 2.mixup

可以模拟遮挡效果

def load_mixup_image_and_boxes(self,index):
        indices = [index] + [random.randint(0, len(self.labels) - 1) for _ in range(1)]
        r_img, _, (r_h, r_w) = load_image(self, indices[1])
        img, _, (h, w) = load_image(self, index)
        mixup_image = cv2.addWeighted(r_img,0.5,img,0.5,0)
        # mixup_image = (r_img + img) / 2
        labels = self.labels[index].copy()
        r_labels = self.labels[indices[1]].copy()
        mixup_labels = np.concatenate((labels,r_labels),0)
        final_labels = mixup_labels.copy()
        final_labels[:, 1] = w * (mixup_labels[:, 1]-mixup_labels[:, 3]/ 2)
        final_labels[:, 2] = h * (mixup_labels[:, 2]-mixup_labels[:, 4]/ 2)
        final_labels[:, 3] = w * (mixup_labels[:, 1]+mixup_labels[:, 3]/ 2)
        final_labels[:, 4] = h * (mixup_labels[:, 2]+mixup_labels[:, 4]/ 2)

        img, labels = random_affine(np.array(mixup_image,dtype=np.uint8), final_labels,
                                      degrees=self.hyp['degrees'],
                                      translate=self.hyp['translate'],
                                      scale=self.hyp['scale'],
                                      shear=self.hyp['shear'],
                                      border=0)  # border to remove
        show(img, labels[:, 1:])
        return img,labels

---

• 3.kmeans 聚类 anchor

---

• 4.边框回归损失（bbox regression Loss）
  
  

---

• 总览

---

• Pseudo-Labelling

---

• Model Ensemble

---

• TTA

补充：数据增强，有时候不能太强，会造成原图极大的变形，导致训练效果很差

---

• OOF

---

• WBF

相比于NMS的操作直接删除的操作，没有利用到重叠度较高的框，而WBF则完事了这个缺点，没有直接删除，而是进行了融合。

---

右边的一般比左边的好

本文地址：https://blog.csdn.net/qq_41375318/article/details/107561914